Github user holdenk commented on a diff in the pull request:
https://github.com/apache/spark/pull/8564#discussion_r39898676
--- Diff:
mllib/src/main/scala/org/apache/spark/ml/regression/LinearRegression.scala ---
@@ -298,11 +298,7 @@ class LinearRegressionModel private[ml] (
*/
// TODO: decide on a good name before exposing to public API
private[regression] def evaluate(dataset: DataFrame):
LinearRegressionSummary = {
- val t = udf { features: Vector => predict(features) }
- val predictionAndObservations = dataset
- .select(col($(labelCol)),
t(col($(featuresCol))).as($(predictionCol)))
-
- new LinearRegressionSummary(predictionAndObservations,
$(predictionCol), $(labelCol))
+ new LinearRegressionSummary(transform(dataset), $(predictionCol),
$(labelCol))
--- End diff --
I'm not really sure handling linear regression without a prediction column
makes a lot of sense, or even explicitly checking the predictionCol here. The
previous version assumed it was set, as do many of the other transformations
(it also has a default value of "prediction"). I think if the user
intentionally clears the prediction column or sets it to an invalid column it
would be better to stick with the existing behaviour and messaging for
consistency across the methods. That being said I will of course defer to your
judgement.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
